Bayesian and frequentist perspectives agree on dynamical dark energy

TL;DR

This study demonstrates that Bayesian and frequentist analyses of dark energy models using DESI and other cosmological data are in strong agreement, supporting the presence of dynamical dark energy rather than a cosmological constant.

Contribution

It validates the consistency between Bayesian and frequentist constraints on dark energy parameters using multiple datasets, and clarifies dataset contributions and internal inconsistencies.

Findings

Bayesian and frequentist constraints on $w_0$, $w_a$ agree well.

Current data is more sensitive to the derivative of $w(z)$ than its mean offset.

Internal dataset inconsistencies under $ m extLambda$ are resolved with a dynamical dark energy model.

Abstract

Baryon acoustic oscillation data from the Dark Energy Spectroscopic Instrument (DESI) show evidence of a deviation from a cosmological constant $\Lambda$ within a Bayesian analysis. In this work, we validate that frequentist constraints from profile likelihoods on the Chevallier-Polarski-Linder parameters $w_0$, $w_a$ are in excellent agreement with the Bayesian constraints when combining with Planck cosmic microwave background, Planck and Atacama Cosmology Telescope lensing, and either Pantheon+ or Dark Energy Survey Y5 supernova data. Further, we assess which datasets drive these constraints by considering the contributions to the $\chi^2$ from the individual datasets. For profile likelihoods of the matter fraction $\Omega_\mathrm{m}$, such an investigation shows internal inconsistencies when assuming $\Lambda$, which are resolved when assuming a $w_0w_a$ dark-energy model. We infer the equations of state $w(z)$ at the pivot redshifts, supporting previous interpretations that current data appears to be more sensitive to the derivative of $w(z)$ rather than a mean offset from $\Lambda$. Thus our frequentist analysis corroborates previous findings on dynamical DE.